Law as language (in contemporary analytic philosophy)1 J.J. Moreso and Samuele Chilovi Suppose that at t the Master says to the Slave !φ; and suppose that the sphere of permissibility just before t contains some worlds, accessible at t, where φ is false at t. Then the sphere must contract to cut those worlds out: at t, and thereafter at least until the next change, none of those worlds are permissible. If the Master changes the sphere in this way by saying !φ, we say that the Master commands that φ. D. Lewis (1979a), ‘A problem about permission’ 1. Introduction According to the approach to ontology that has been pursued within the analytic tradition in philosophy at least since Quine (1948), the ontological question about some candidate entity x can be stated in English with three words – ‘does x exist?’ – and answered with one.2 Since in this tradition ontological questions are conceived as existence questions, to inquire into the ontology of something is first and foremost to ask whether that thing exists. Given that in principle one may ask the ontological question with respect to anything, it is natural to think of the subject matter of This paper is the translation of the penultimate draft of ‘Il diritto come linguaggio (nella filosofia analitica contemporanea)’, to appear in G. Bongiovanni, G. Pino, C. Roversi (eds.), Che cosa è il diritto. Ontologie e concezioni del giuridico, Giappichelli, Torino, 2016. Section 4 of the original version has been omitted, and the published version is significantly different; for in-depth study we recommend readers to consult the official version. The text is conceived as a textbook introduction to the subject for law students. 2 Actually, according to Quine the question of ontology should be stated in quantificational idiom; but then, since for him existence is the same as being, and “to be is to be the value of a bound variable” (Quine 1939: 708), it is not inaccurate to ascribe to him the view that ontological questions are existence questions. 1 1 legal ontology as concerned with the issue of whether law exists. Although we do not wish to understate either the difficulty or the depth of this question, for the purposes of this paper we will work on the assumption that it received an affirmative answer, and will focus rather on tackling a different, although closely related, one. We may call the question we’ll be dealing with ‘metaphysical’. To ask the metaphysical question about some (candidate) entity x is to ask of x what it is or, more precisely, what in the nature of x makes it what it is.3 In this respect, the main thesis that will be articulated is that it lies in the nature of law that part of it bears to language a very special relation, a relation that holds in virtue of the fact that laws are, essentially, contents. The connection that ties law with language can be schematically articulated in three steps. First, the law of a given jurisdiction at a given time can be thought of as the legal system of that jurisdiction at that time;4 second, legal systems can be viewed as sets of laws;5 third, laws can in turn be modeled as contents that have the property of being legally valid (in a jurisdiction, at a time). If laws are contents, presumably they are the contents of certain linguistic items – namely, texts or utterances (made up) of written or spoken words, and arranged through syntactic rules. Hence, part of what constitutes what laws are would seem to bear to language the very same relation that contents (pieces of information) bear to the linguistic expressions they are the contents of. In legal jargon, the well-formed linguistic expressions that laws are related to are called ‘legal provisions’, although in a narrower sense of this phrase not all laws need to be related to legal provisions in this way. Thus, consider the following provision: (1) Smoking is forbidden. (1) is a linguistic item – specifically, a sentence of English. To say that (1) is a legal provision is to say that it is a sentence that is contained in a legal source – a text that went through a We write ‘candidate’ because one may ask the metaphysical question before the ontological question; in this case, it would take a slightly different form: ‘what would x be if it existed?’. We set aside the complicated issue of the relation, and order of priority, between ontology and metaphysics, which is not the topic of this paper. 4 We use the term ‘jurisdiction’ in a non-technical sense to refer to a (legally or socially defined) location or region of space. 5 We use the count noun phrase ‘a law’ and the bare plural ‘laws’ in a non-technical sense to refer to a legal norm of any kind. We use the mass noun phrase ‘the law’ to refer to a legal system, or set of laws. 3 2 lawmaking process which ended successfully. What content does (1) express, that is to say – if our working hypothesis is correct – what is the content of the law that is related to (1) in the appropriate way? Roughly speaking: that smoking is forbidden. Although this answer will appear unsatisfying to many, we shall first try to show why it is in fact more informative than it appears, and then illustrate three ways of analyzing that-clauses like the one above that are prominent in the literature on propositions. This, in turn, should show why the preceding remarks are but the tip of the iceberg within a full-blooded metaphysical account of law. Contents are the objects of intentional states like belief or acceptance, they are the sort of things that can be complied with or violated, and the possible bearers of the property of legal validity. As to the ontological category to which they belong, they are arguably a kind of abstract object, along with numbers, sets and values, and unlike tables, chairs and atoms. While engaging with a classical tradition according to which abstract objects are thought to exist outside the spatial and temporal dimension, we will try to reconcile this view with the intuition that law’s existence in fact depends on human acts and mental states. In particular, we will try to achieve this result by drawing on the impact that these acts and states have on the contents’ acquisition and loss of validity. In connection to this, we will see how the notion of lawmaking can be usefully embedded in the framework provided by speech act theory in order for it to be informatively explicated, and will review the main accounts of the nature of propositional content. Finally, we will focus on the notion of validity and distinguish it from that of existence, so as to dispel a few misunderstandings that surround them, and to introduce the reader to a central question regarding the ontology of law. 2. Lawmaking 2.1 Speech act theory and levels of meaning Speech act theory is the branch of pragmatics (the study of language use) that originated with the work of J.L. Austin (1962). The main theme that runs through this line of research – that people can do things with words – is developed by studying the ways in which agents can perform acts in 3 using language, and the acts that are thereby performed. When someone utters a sentence, the act s/he performs can normally be split and analyzed in three components, that Austin famously named ‘locutionary’, ‘illocutionary’, and ‘perlocutionary’. Thus, for instance, with respect to an utterance of ‘Rome is in Italy’ in a particular context, it is possible to identify (i) the locutionary act of uttering a sentence with a definite sense and reference – that Rome is in Italy, and the truthvalue true respectively; (ii) the illocutionary act done in uttering it – in this case, an assertion; and (iii) the perlocutionary act that was performed by uttering it – trying to persuade the intended audience, or to cause in it a corresponding belief, for instance. The perlocutionary level of the utterance lumps together a variety of nonconventional and irregular effects that the utterer may seek to achieve through her act, and captures her attempt to produce in the hearer a certain attitude or to get him to behave in a certain way. These effects on the hearer’s actions and attitudes include such diverse things as persuading, alarming, convincing, scaring, inspiring, and many others. Of higher interest to the study of communication are the notions of locutionary and illocutionary act. The content of a locutionary act is widely taken to be a proposition,6 and in particular the proposition that corresponds to the level of content of the utterance that is technically called ‘what is said’. Let us explicate this notion as inscribed within the levels of meaning that an utterance might have. First, one can identify the literal meaning that is conventionally encoded by the uttered sentence; this is a function of the meanings of the subsentential expressions that make it up, and of the syntactic rules through which they are arranged. In simple cases, this conventional meaning manages to achieve the status of a complete proposition, evaluable for truth and falsity. For instance, the English sentence ‘snow is white’ means that snow is white; and the proposition that snow is white is true if, and only if, snow is in fact white. However, in cases in which the sentence contains a pure indexical element (a word such as ‘I’, ‘here’, and ‘now’), since the reference of these expressions changes depending on the context of utterance, sentence meaning alone is no 6 For an explication of the notion of proposition, see section 3. 4 longer able to determine a truth-evaluable content. For instance, if Miguel utters the sentence ‘I am Spanish’, literal meaning only tells us (roughly) that the sentence is true just in case the speaker is Spanish, and this information is insufficient for us to know the conditions under which it is true. In order to know this, we need to be supplied with contextual information – in this case, about the identity of the utterer. Once we have that, we can use it to retrieve the semantic content of the sentence relative to the context of utterance, which constitutes the second level of content. The semantic content of ‘I am Spanish’, relative to a context in which Miguel is the speaker, is that Miguel is Spanish – a proposition which is either true or false, hence endowed with definite truth-conditions. At the next level, we find sentences that contain expressions which, in order to be supplied with semantic values, need to be processed through mechanisms of assignment that are not as automatic as those that are involved in uses of pure indexicals. Those include the pronouns ‘he’ and ‘she’ when used anaphorically, demonstrative expressions (‘this’, ‘that’), possessive constructions (‘Miguel’s book’), quantifiers (‘every’, ‘any’, ‘all’, ‘some’), and sentences that are syntactically complete but semantically incomplete, such as ‘I’m ready’. Once disambiguation, reference fixing, and other kinds of semantic value assignment procedures which are needed to handle these cases have been carried out, we get to the level of propositional content that we previously called ‘what is said’. By uttering ‘John’s book is wonderful’, Miguel said that the book that was written (edited, published, etc. …) by John is wonderful. Similarly, by uttering ‘I am ready’ Miguel said that he was ready to go to school (to the party, to have lunch, and so on). Finally, we encounter the level of content that was called by P. Grice (1989) ‘implicature’. An implicature is a proposition that is conveyed by the speaker implicitly, over and above what the sentence literally means and what the utterer says. As such, implicatures constitute a part of speaker meaning, and hence are determined by the speaker’s communicative intentions. However, this is not to say that their behavior is utterly unpredictable, as in fact Grice himself laid down a 5 number of maxims through which their production could be abductively calculated.7 These maxims are all subject to a general principle of cooperation that guides the conversation, and of which they provide specific ramifications.8 Inasmuch as the agents involved in the linguistic exchange are expected (and expect one another) to comply with them when making a contribution to the conversation, qua presumptions on the speaker’s communicative intentions, the maxims are relied upon by the hearer in inferring the implicatures that the speaker produced. To take up a previous example, by saying ‘John’s book is wonderful’ to Carlos, in the appropriate circumstances Miguel might have implicated, i.e., that Carlos should also read the book. Examples and complexities abound in each of the themes we touched upon, but the rough characterization we provided should be sufficient to understand the discussion that follows. Having introduced the notion of a locutionary act and the levels of utterance meaning, we now turn to an investigation of the notion that plays the most central role within the study of speech acts, that of illocutionary force. 2.2 Illocutionary acts As we saw in the previous section, in uttering a given sentence the speaker may perform an illocutionary act. In saying that Rome is in Italy, Miguel made an assertion; in telling Carlos to shut the door, he issued a command; in asking Marc whether the lawgiver is wise, he made a question – just to mention a few. As Garcia-Carpintero (2004) has pointed out, ‘illocutionary force’ is ambiguous between a type of illocutionary force, and the combination of one of them with a content; we shall use the phrase with its former meaning. 7 The maxims of conversation are: A) quality: try to make your contribution one that is true; B) quantity: 1. Make your contribution as informative as is required; 2. Do not make your contribution more informative than is required; C) relation: be relevant; D) manner: be perspicuous. (Grice 1989). 8 The cooperative principle reads as follows: make your conversational contribution such as is required, at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged. (Grice 1989: 26). 6 Necessarily, speech acts are acts with contents. Acts and their contents, Frege maintained,9 are distinct, both in the sense that neither is a part of the other, and in the sense that neither fully determines the other. The distinction between force and content, although not universally accepted,10 still remains a central tenet of contemporary speech act theory (see Searle 1969). Just as the content of an occurrence of (say) ‘you’ll be more punctual in the future’ does not determine whether a prediction or a threat has been made, the mere fact that someone has made an assertion does not determine what has been asserted.11 Furthermore, the propositional content of the speech act can not only be isolated with respect to its force, but also from grammatical mood. For instance, the propositional content that the door is shut is shared by ‘the door is shut’, ‘is the door shut?’, and ‘shut the door!’.12 Thirdly, as we saw from the fact that a sentence in the indicative mood might be used to issue a threat, it seems that propositional content fails to determine the force of the act also in combination with grammatical mood. Accounts of illocutionary acts typically have both a taxonomical and an explanatory aim. On the taxonomical side, they try to classify an indefinite number of forces by fitting them in a small class of general types. In the literature, two are the catalogues that have enjoyed greater success. John Searle (1985) divides them among assertives, directives, commissives, expressives, and declarations. Bach and Harnish (1979) partition them into constatives, directives, commissives, acknowledgments, effectives, and verdictives. However, these two groupings are similar in that one important criterion through which they classify types of force is their direction of fit, of which two species can be isolated. When the act bears a direction from word to world (paradigmatically in the case of assertives and constatives), the content of the words aims at fitting the world, an aim that is realized to the extent that the representation of it that they give is correct. By contrast, speech acts with an opposite direction (paradigmatically directives and commissives) aim at matching the world to the content of their words. In the example that Anscombe (1957) 9 But see Hanks (2007) for a criticism of the standard historical reading of Frege. See Hanks (2007) for a forceful criticism of it. 11 The example is drawn from Green (2015). 12 See Green (2015). 10 7 famously uses to illustrate the distinction, a wife gives to her husband a shopping list, and invites him to go to the supermarket to buy the items which are listed therein. While filling the shopping bag with those items, the husband is fitting the content of the bag to the content of the list: he is trying to adapt the world to what the list says. Meanwhile, a detective is following him, compiling a paper where she takes notice of all the items that the husband is buying; in doing so, the detective is trying to fit the content of the list to that of the husband’s bag. Therefore, if the detective succeeds in her task, there will be two lists containing the same items, and thus having the same contents, but with opposite directions of fit. Apart from this brief digression, here we won’t dwell on all the features of each of the categories that make up the taxonomies of illocutionary acts, although we will look at some of them when exploring the type to which lawmaking belongs. The explanatory aspect is more relevant to our present project. The purpose of this part of the theory lies in explaining what any given speech act essentially is, in uncovering those facts (or other sorts of things) in virtue of which some utterance counts as the performance of the specific illocutionary act that it is. In this respect, three views can be highlighted. The first, originally defended by Austin (1962) and later developed by Searle (1969), appeals to the role of conventions in characterizing illocutionary force. In the words of Austin (1962: 14), the thesis is that for any force ‘there must exist an accepted conventional procedure having a certain conventional effect’. This view is vulnerable to two objections. As many authors have emphasized, Austin’s analysis focuses especially on acts whose performance and success requires institutional settings, and it overlooks a wide range of cases where this does not seem to be the case. Secondly, since he left the notion of convention unexplained, the account is incomplete in another respect. Searle’s theory can be read as an attempt to fix these two shortcomings. To account for non-conventional acts, he weakens Austin’s view by allowing that they could be explained within an intentionalist framework. To flesh out the notion of convention, he makes use of Rawls’s (1955) notion of a constitutive rule – which was to be found, in 8 application to linguistic rules, also in Midgley (1959). Those are rules that paradigmatically take the form ‘X counts as Y in context C’, and that impose a new status function (the one that is assigned to the ‘Y’ term) on a pre-existing entity. Thus, according to Searle (1969: 24), ‘speech acts are acts characteristically performed by uttering sentences in accordance with these sets of constitutive rules’. This view has been recently modified and developed by theorists (for instance, Williamson 1996) who take speech acts to be moves in language games that are defined by constitutive rules. The second way to go about answering the explanatory question was initially pursued by Strawson (1964), and subsequently refined by Bach and Harnish (1979). The latter writers split the main illocutionary types in two broad categories: those whose force is regarded as a natural phenomenon consisting in the speaker’s expression of a certain attitude, and those where conventions play a crucial role. On their taxonomy, constatives, directives, commissives and acknowledgments fall within the former category, whereas effectives and verdictives belong to the latter. To illustrate how the theory works in connection with the acts of the first kind, let us briefly mention the authors’ treatment of assertion. Assertion is defined in terms of the expression of belief through the following two-clauses (Bach and Harnish 1979: 42): in uttering a sentence e, a speaker S asserts that p iff S expresses (i) the belief that p, and (ii) the intention that the hearer believe that p. Attitude expression is then defined by reference to a Gricean type of reflexive intention, whereas to express an attitude is to R-intend the hearer to take the speaker’s utterance as a reason to think that the speaker has that attitude, and to ‘R-intend’ some effect is in turn to intend to bring it about by means of the hearer’s recognition of this very intention. By contrast, in the other type of acts conventions replace R-intentions in determining what kind of act an utterance counts as. In line with Searle, the authors take conventions to be count-as rules, and, in the same spirit of the social ontology propounded by him, both effectives and 9 verdictives are said to be capable of affecting institutional states of affairs. Conventions make it the case that utterances count as acts of certain types, while the mutual beliefs shared by the members of the relevant community play a crucial role in establishing conventions. This mechanism is schematized thus (Bach and Harnish 1979: 109): An act of a certain sort A (in a context C) is a convention for D-ing in a group G if and only if: (i) it is mutually believed in G that whenever a member of G does A in C, he is D-ing; (ii) A in C counts as D-ing only because it is mutually believed in G to count as such. The third approach is the so-called ‘dynamic’ one, first put forward by Stalnaker (1999, 1978) in his work on assertion, but also traceable to Lewis’ (1979a) elaboration of the language game of commanding and permitting set forth in ‘A problem about permission’ (see the quotation at the beginning of the paper). This approach was pursued by those philosophers who came to be interested in speech act theory through their work on formal semantics when dealing with contextsensitive expressions. This line of research was urged by the search for a formal modeling of context, which was designed as a tool to capture in a systematic and rigorous way the context’s contribution to the truth-conditions of statements which are sensitive to it. This inquiry in turn led to the elaboration of a formal understanding of conversation and of the linguistic contributions that participants engaged in it can make. In this framework, a conversational context is conceived of either as the set of propositions that are shared (presupposed) by the speakers at a given time in the ongoing conversation – the common ground – or, equivalently, as the set of worlds that are compatible with all such presuppositions – the context set. This technical machinery (combined with a possible worlds treatment of propositional content) then allows the speakers’ speech acts to be understood in terms of the ways in which they aim to alter the conversational context.13 Under a certain interpretation of Stalnaker’s work (for instance, the one advocated by MacFarlane 2011), This is the line of reasoning that was developed by Lewis in ‘Scorekeeping in a language game’ (1979b), and that he used to analyze the speech acts of commanding and permitting (see the quotation at the outset of the paper). 13 10 he defines the speech act of asserting in terms of the essential effect that it has on the context, whereas this peculiar effect consists in proposing14 to add new information to the common ground. Since this addition poses further requirements on the way the world must be if all the speakers’ presuppositions are to come out as true, the making of a statement typically determines a contraction of the context set, for the worlds that are incompatible with the newly added content are to be removed from it when the statement is made. Recently, some theorists have also tried to use dynamic theories to offer an explication of directive force. This has been done by introducing a notion analogous to that of context and by modeling directive speech acts in terms of their capacity of updating it. Portner (2007), for instance, posits in place of context a To-Do-List, namely a complete representation of the set of states of affairs that the addressee must bring about, or of the properties that he ought to possess. This set of obligations can then be altered as the conversation evolves by those engaged in it, and directive force is defined as the kind of act that can produce such an alteration.15 Now that we have at our disposal a taxonomy for classifying forces and some accounts that explain them, we can use them to account for the nature of lawmaking. 2.3 Lawmaking as an illocutionary type Pre-theoretically, it seems intuitive to think of lawmaking primarily as a verbal business. A good deal of a lawmaker’s job typically consists in drafting texts and putting them up for vote, so that crafting a legal material with the appropriate wording appears to be a central ingredient of our ordinary legal practices. For this reason, it would be good to have an in-depth understanding of this phenomenon. Speech act theory, we think, affords the resources to do so, roughly because law-creation happens to be a paradigmatic case of how one can do things with words, wherein saying something makes it so. 14 15 And in adding this information, if there is accommodation. For a nice and useful introduction to these topics, see Harris (2014). 11 Although law-creation involves the performance of speech acts, it is also true that (i) normally the mere performance of utterances is not enough to that end, and that (ii) it is a nontrivial question just what kind of speech act law-making is. The following remarks can be read as an attempt to address these two concerns. The first point is clear from the fact that not everybody can make law, and that no one can, unless he or she is in the right contextual settings, doing the things that are appropriate given that setting. In this respect, although they share the same direction of fit, lawmaking and commanding are very different, since the latter enjoys a much wider range of occurrences. This remark points to the conventional element that characterizes lawmaking. This aspect is accounted for both in Searle’s general account, and in Bach and Harnish’s treatment of effectives and verdictives. In particular, lawmaking seems to fit pretty well in Searle’s declarations, for, as he says (Searle 1985: 16-17): ‘it is the defining characteristic of this class that the successful performance of one of its members brings about the correspondence between the propositional content and reality. (…) The performance of a declaration brings about a fit by its very successful performance’. And for this to be possible, there has to be a system of constitutive rules in addition to the constitutive rules of language. Independently of how this works in detail, what is important to notice is the role of conventions and the fact that the specific import of a declaration is that of creating an institutional state of affairs. In the special case of lawmaking, this import is modeled by Searle as the effect consisting in the fact that the propositional content of the speech act becomes law. Very closely related to this are the remarks that Bach and Harnish (1979: 110-115) offer about effectives. These are acts that are said to be: 1. fact-producing, i.e. they are such that, when issued by the right person under the right circumstances, make it the case that such-and-such; 2. necessarily conventional, i.e. governed by constitutive rules that exist in virtue of mutual beliefs shared by the members of a community; 12 3. such that their nature lies in the fact that they effect – produce or alter – institutional states of affairs. Two important points have to be drawn from these remarks. The first is that, when it comes to considering acts of the type that we are interested in, Searle’s and Bach and Harnish’s accounts converge, both acknowledging the essential role of conventions in act-type constitution. The second is that, although both accounts are traditionally characterized in opposition to dynamic ones, they end up defining lawmaking in terms of the essential effects that an utterance that instantiates it brings about, in keeping with the main tenet of the dynamic proposals.16 After analysis, it turns out that these three approaches tend to overlap when dealing with the object of our inquiry. In light of the preceding discussion, we put forward the following definition: an appropriate agent A law-makes a content p in a jurisdiction j in uttering (inscribing) a sentence e iff A updates the legal system L of j by adding p to it. According to this definition, to make a law is essentially to alter the legal system by updating it through the addition of one or more of its contents. The reason why legal systems come into play is that they are able to perform a useful theoretical job. On the one hand, they can be employed to characterize the law of a jurisdiction at a time as the set of the laws it has (at that time); on the other hand, they can also play a semantically perspicuous role, for they can serve as the counterpart of Stalnakerian common grounds in the context of non-assertoric speech acts, and can thus be used to represent the informational contribution that successful lawmaking acts make to the content of the law.17 It could also be argued that at least in some cases legal provisions are used to perform two illocutionary acts at the same time. Especially when legal provisions take the form of deontic 16 Searle (1985: 28) mentions the act of promulgating a law as an example of a declaration, and Bach and Harnish (1979: 114) take law-making to be a species of effectives. 17 Semantically, a legal system is therefore a superset of Kratzer’s (1977) conversational background for deontic modal expressions. 13 modal sentences, utterances of them may – in addition to constituting lawmaking acts – also bear directive force. For instance, a (performative) utterance of (1) may not only have the effect of prohibiting a certain behaviour, but also be intended as a guide for the conduct of those to whom the resulting rule is targeted. This position gains strong support from the observation that one of the essential features of law plausibly lies in the fact that it aims at guiding people’s behaviour, perhaps by providing them with reasons for action.18 And while it is far from obvious that law’s normativity (its ability to bestow people with reasons for actions) could be explained via its directive force, it is rather natural to think that its conduct-guiding role could find there a natural explication of its source. As a side note, we also point out that the multiple force hypothesis seems to have been embraced by Searle (1985: 28) where he says: ‘Promulgating a law has both a declarational status (the propositional content becomes law) and a directive status (the law is directive in intent)’. So far, we have dealt with the dynamics of law creation, outlining a way for explaining how uses of language can determine what the law says. Now, we wish to explore the first of the two elements of which any law is composed. Exposing the salient theoretical options in point takes us into an investigation of the nature and structure of content. 3. The nature and structure of content 3.1 The ontology of propositional content Sentences of natural languages encode pieces of information. This is the information one has to get a grip on in order to understand the sentence, the interpretation that the sentence is assigned, its meaning. In more theory-laden terms, natural language semantics makes use of the notion of semantic content in order to represent the semantic value of the well-formed formulas that emerge, and can be calculated, from the semantic values of their parts and their mode of syntactic 18 Alternatively, one could try to explain this central feature of law by regarding the directive dimension of lawmaking as pertaining to the perlocutionary level. 14 combination. Meaning, on this approach, is regarded as an entity.19 Within an extensional framework, semantic values are called ‘extensions’; the extension of a sentence is a truth value, that of a singular term an individual, and that of a predicate a set, or the characteristic function of the corresponding set. More complex semantic values are assigned to n>1-place predicates, connectives, quantifiers, determiners and prepositions by constructing complex functions built out of individuals, truth values, and simpler functions. Within an intensional framework, all semantic values are functions from worlds (or more complex indices) to extensions, and the semantic value of a sentence is called ‘proposition’. A crucial question that one may be willing to ask is: why can’t we get along by appealing to sentences and the simple expressions that compose them alone, and dispense with propositions altogether? One popular answer to this question has it that propositions earn their living by performing the theoretical jobs that they are called on to fulfill in our theories. This methodological stance can already be found in Lewis (1970: 195), where he writes: ‘In order to say what a meaning is, we may first ask what a meaning does, and then find something that does that’. In keeping with this suggestion, let us move on to see the theoretical roles that propositions are called on to play. The first motivation appeals to the pre-theoretical fact that one sentence can have different contents, and different sentences can have the same content. The English sentence ‘I am Italian’ means that José Juan is Italian as uttered by José Juan, and that Samuele is Italian when uttered by Samuele. On the other hand, not only the Italian sentence ‘la neve è bianca’ and the Spanish sentence ‘la nieve es blanca’ appear to have the same meaning, but the English sentences ‘Giovanni loves Giulia’ and ‘Giulia is loved by Giovanni’ seem to share it as well. Therefore, the argument goes, sentences and their meanings cannot be the same thing. A sentence is a grammatical entity, an ordered sequence of symbols that belongs to a particular language and has the semantic properties it does in virtue of the linguistic practices of a certain community. By 19 More on which kind of entity is said in the next sub-section. 15 contrast, a proposition is an abstract entity tied to sentences across different languages, and is thus language and mind independent. Another theoretical job that propositions are thought to perform is that of being the primary bearers of truth values. Although we ordinarily say of sentences that they are true or false, which way they are always depends both on how the world is and on the piece of information they encode. So we can only say that a given sentence is true (false) if the content it expresses is (not). Many authors have accordingly accounted for this fact by holding that sentences possess truth values derivatively, in virtue of standing in a certain relation to the contents to which they are so related. The third motivation has to do with the modal (and other sorts of) attributes that we ordinarily ascribe to them. Commonly, we say of a truth that it is necessary, possible, or contingent; and this, intuitively, can be paraphrased by saying that the corresponding proposition is necessarily (possibly, …) true. Further, for those who think (as we do) that rules are contents, it will be natural to understand what we say when we say that a rule is accepted, complied with, or valid, as saying that the corresponding contents possess the relevant attributes. The fourth job that propositions are assigned is related to their role in our linguistic practices and mental lives. As we saw in dealing with the dynamics of language use and the study of illocutions, there are many things that we can do with words: we use them to assert, doubt, suggest, hypothesize, deduce, and so on. Now, the question that arises is: what are the things that we assert (doubt, …) when we perform such acts? If we allow propositions in our ontology, this question admits of a natural and simple answer: propositions are such things. In addition to this, if we follow this route it is also easy to get a semantic interpretation of attitude verbs such as those mentioned above. ‘Doubt’ (‘assert’, …) would express a two-place relation between agents and contents, and so the fact that someone doubts something will be analyzed as her standing in the doubting relation to the proposition she doubts. Analogous to this is the account that we would get of the semantics of terms for intentional mental states such as ‘belief’, ‘judgment’, or 16 ‘entertainment’. Like propositional attitude verbs, the expressions denoting these states could also be taken to express relations between individuals and contents. Accordingly, believing something would be analyzed as the standing by an agent in the appropriate cognitive relation to the proposition that she believes. Finally, theories of propositions have proved to be useful in natural language semantics for understanding the meaning of ‘that’ clauses. These are clauses that one can easily construct by prefixing a declarative sentence with the word ‘that’. For instance, if you take the sentence ‘Barcelona is in Spain’, and prefix it with ‘that’, you get ‘that Barcelona is in Spain’. This expression, on the face of it, looks like a singular term whose function is to enable us to refer to a certain thing and to predicate of it certain attributes. To remain on the example, that Barcelona is in Spain may be said to be surprising, boring, fortunate, true, and so on. A related point here is that if one takes that-clauses like these to refer to propositions, one is able to explain in an elegant and simple way the intuitive validity of arguments of the following form: P1. Miguel believes everything Paul says; P2. Paul said that Barcelona is in Spain; C. Miguel believes that Barcelona is in Spain. For if that-clauses really denote propositions, one can formalize the argument above in a way that accounts for its intuitive validity in the following way: P1. For all x, if Paul says x then Miguel believes x; P2. Paul said p C. Miguel believes p. In conclusion, these are the main considerations that advocates of propositions appeal to in justifying their appeal to this notion. Whether these arguments are conclusive is an open question, 17 but we do seem to have enough reason to work on the hypothesis that propositions exist. Let us then take a look at what they might be.20 3.2 Propositions as structured entities Both Frege and Russell thought that propositions were needed qua sentence meanings, primary bearers of truth and falsity, and objects of intentional attitudes and acts. Furthermore, both of them thought of propositions as structured complexes composed by some kind of constituents. This idea was suggested by the observation that sentences are complex entities with a syntactical structure and composed by simpler (sub-sentential) elements. Since the words and phrases that combine to construct sentences were independently thought to possess meaning, they were naturally led to conceive of propositions as analogous complexes composed of the meanings of the expressions making up the sentences by which they are expressed. For Frege and Russell, propositions are essentially endowed with truth-conditions, which enable them to represent reality as being some way or other. The representational property of having truth-conditions was taken to be not only essential to them, but also independent of any human act or mental capacity. Being abstract entities, they were thought to lie outside space and time, and to exist independently of being grasped or entertained by anyone. However, here is where the similarities between the two thinkers come to an end. For those who regard propositions as composite objects, the main challenge lies in specifying the nature of the constituents that make them up, and of the relation that binds them together. Frege and Russell held views that differ on both of these topics. 3.2.1 The Fregean view One of Frege’s (1892/1960) most important contributions to the philosophy of language is his distinction between sense and reference. He was struck by the difference in cognitive significance 20 For an excellent introduction to these themes, see King, Soames, and Speaks (2014). 18 between identity statements such as ‘a = a’ and ‘a = b’. Whereas the former says of an object that it is self-identical, and thus can be known to be true a priori, the latter might be highly non-trivial, as it may take some empirical investigation to know whether it is true. For instance, in the example that he gives, coming to know the truth of a sentence like ‘the morning star is the morning star’ requires no empirical inquiry, whereas that of ‘the morning star is the evening star’ does. The challenge that he thought this poses to the theory of meaning is the following. If one takes the meaning of a (singular) term to be just the object that is denoted by it, one has a hard time explaining how these two statements could be any different, since both ‘the morning star’ and ‘the evening star’ refer to the same object, namely the planet Venus. This line of reasoning led him to posit two aspects to linguistic meaning: sense and reference. Reference is the thing that an expression refers to: an object in the case of singular terms, a concept or function in the case of general terms, and a truth value for sentences. A sense is the objective ‘mode of presentation’ that an expression gives of its referent, and what determines its denotation: (usually expressible by) a definite description in the case of singular terms, and a thought or proposition in the case of sentences. This allowed him to say that a sentence like ‘a = b’, where the object referred to by ‘a’ and ‘b’ is presented in different ways, has a different sense from that which is associated with ‘a = a’. And this enabled him to explain the difference in cognitive significance between the two statements. Frege also subscribed to the thesis of the compositionality of meaning. This is the view that the meaning of a complex expression is a function of the meaning of its parts and the rules of syntactic composition. Within his framework, this idea was to be implemented as the thesis that the sense or reference of a sentence is a function of the senses or referents of its parts and of how they are arranged. At times (Frege 1892: 54), he also expressed the stronger view, typically associated with structured accounts of content, that the meaning of a sentence is composed of the meanings of its constituent expressions. Given this further commitment, it was natural for him to proceed towards an inquiry of how meanings are held together, and he looked for the source of 19 this link within the constituents themselves. More specifically, he thought of certain expressions – predicates – as denoting entities which are unsaturated, or incomplete, and of others – sentences and singular terms – as saturate. In particular, he thought that general terms denote functions that take arguments and yield as values truth-values. The predicate ‘is Italian’, for instance, is incomplete, and denotes a function that returns the value true for any Italian individual, and false for any other. Saturation, then, is the mechanism through which predicates are completed, and through which the semantic value of a complex expression is determined on the basis of its parts. Given that unsaturated expressions have functions as meanings, semantic composition is nowadays generally thought to be functional application.21 Thus, if the semantic value of (say) ‘is Italian’ is given as argument Giulia, and if Giulia is Italian, the function will assign as value true and the sentence ‘Giulia is Italian’ will hence come out as true. In conclusion, for Frege one can only explain how propositional constituents are bound together by recognizing the existence of unsaturated entities and of the process of saturation: a proposition is unified through the application of a function to its argument. 3.2.2 The Russellian view Russell (1903) held that the meanings of sub-sentential expressions are their referents, and that those are objects, properties, and relations (for singular terms, monadic and polyadic predicates respectively). Further, he also viewed propositions as composed by those very things. For instance, the proposition that Socrates is human is made up by Socrates and the property of humanity. This proposition predicates the property being human of Socrates, and hence has truthconditions: it is true if and only if Socrates is, in fact, human. As to the question of what binds the 21 See Heim and Kratzer (1998), who interpret semantic composition as functional application, and take this to be one of the fundamental insights and legacies of Frege’s work. Here is what Frege (1891) says, as quoted by them (Heim and Kratzer 1998: 3): ‘Statements in general , just like equations or inequalities or expressions in Analysis, can be imagined to be split up into two parts; one complete in itself, and the other in need of supplementation, or "unsaturated." Thus, e.g. , we split up the sentence "Caesar conquered Gaul" into "Caesar" and "conquered Gaul." The second part is "unsaturated" – it contains an empty place; only when this place is filled up with a proper name, or with an expression that replaces a proper name, does a complete sense appear. Here too I give the name "function" to what this "unsaturated" part stands for. In this case the argument is Caesar.’ 20 propositional constituents together, Russell thought that the answer was to be found in the semantic contribution of the sentence’s verb. In the proposition expressed by ‘Socrates is human’, Socrates is said to stand in the relation denoted by ‘is’ to the property of humanity. Both the view of Frege and Russell suffer from serious difficulties, which recent approaches have tried to overcome by modifying substantial aspects thereof. However, we reviewed them in order to appreciate the main metaphysical features and challenges that theories of content that conceive of propositions as structured entities have and face. These views have the burden of giving an account of the nature of propositional constituents, and of the relation of propositional constituency that holds them together. Both of these themes have currently received widespread attention by linguists and philosophers. 3.3 Propositions as sets of possible worlds Shortly after the revival of modal logic through the work of Carnap (1947), Marcus (1946), Hintikka (1957), Kripke (1959) and others, formal semanticists began to develop semantic frameworks designed to take care of intensional phenomena while remaining in the tradition of truth-conditional semantics inaugurated by Frege. Very briefly, intensional expressions are those that determine the failure of the principle of substitutability of co-referential expressions salva veritate with respect to sentences that contain them, while intensional contexts are the sentences that are embedded under intensional expressions.22 Since this principle is closely tied to the principle of compositionality, such failures were taken very seriously by those working in this area, and urged them to look for a way of amending the semantics that could save it. This problem led to the development of possible world semantics, of which we earlier saw what semantic values it employs. This system includes a special compositionality principle that becomes operative whenever the presence of an intensional expression triggers it, and specifies types of semantic values that are apt to be fed into it. 22 The substitutability principle in extensional semantics says that any two expression that are co-extensional can be substituted for one another in any sentence salva veritate. 21 Within this framework, propositions are viewed as characteristic functions of sets of possible worlds; while this may appear to be somewhat difficult to grasp, once one understands what it means, the resulting idea is in fact quite intuitive. Stalnaker (1970: 381) puts it thus: The explication of proposition given in formal semantics is based on a very homely intuition: when a statement is made, two things go into determining whether it is true or false. First, what did the statement say: what proposition was asserted? Second, what is the world like: does what was said correspond to it? What, we may ask, must a proposition be in order that this simple account be correct? It must be a rule, or a function, taking us from the way the world is into a truth value. But since our ideas about how the world is change, and since we may wish to consider the statement relative to hypothetical and imaginary situations, we want a function taking not just the actual state of the world, but various possible states of the world into truth values. There seems to be much of intuitive to the thought that propositions are sets of possible worlds, even if one neglects how useful the idea has proved in the rigorous study of meaning. Here we are not concerned so much with the semantical aspects of the matter, as rather with the metaphysical ones. Within the account we are considering, those boil down to understanding the nature of functions, sets, and possible worlds. Let us set aside the first two, and focus instead on possible worlds. 3.3.1 Possible worlds The central metaphysical question about possible worlds is: what are they? Most of the views that have sought an answer to this question can be grouped in two broad families, that we call ‘concrete modal realism’ and ‘abstract modal realism’. 3.3.1.1 Concrete modal realism The author who first formulated this view, and who is by far the one who gave it the best articulation and defense, is David Lewis. Lewis (1986) forcefully embraces an ontology of possible worlds because he regards them as indispensable in performing a host of theoretical jobs. Apart from appealing to them in giving an account of propositions, he also deploys them in 22 explicating a variety of other intensional entities. In metaphysics, they serve to characterize in clear and rigorous terms properties and relations, accordingly viewed as functions from worlds to sets of individuals, and from worlds to sets of n-tuples of individuals respectively. In semantics, they are used to provide with truth conditions statements containing intensional expressions, where modal adverbs and operators are treated as quantifiers whose variables range over worlds: a truth is taken to be necessary if it is true in all possible worlds, and possible if it is true in some possible world. In keeping with the treatment of modal statements given in possible world semantics, Lewis regards the modal notions of possibility and necessity not as primitive, but rather as reducible to worlds and their objects. Roughly,23 when we say (2) Obama could have lost the elections, what makes this sentence true – as it intuitively is – is that there is a world where Obama lost. And when we say that no triangle could fail to have three sides, what we say is that in all possible world every triangle does in fact have three sides. While the phrase ‘the actual world’ is understood very broadly, as the entire universe we live in, Lewis also thinks that in order to know what a possible world is like, we should just look around us: other worlds are just like our own in kind, and exist in the same way that ours does. More precisely, possible worlds are defined as maximal connected objects, where an object o is connected if any two of its parts bear a spatiotemporal relation to one another, and o is maximal if it has no part that is spatiotemporally related with something that is not one of its parts.24 Thus, a characteristic feature of Lewis’ system is that the actual world enjoys no ontological privilege with respect to any other; it is on a par with all of them with respect to both being and kind. Semantically, this is reflected by the fact that ‘actual’ is treated as an indexical term, just like ‘here’. Accordingly, each world and person is actual from its own perspective, and we are no more actual from ours than an alien individual is from hers. Since on this view different worlds bear no spatiotemporal connection to one another, 23 24 We will shortly see that for Lewis things are not exactly how we are presenting them now. We follow Menzel (2015) in this presentation. 23 no causal link can intervene between them, and none of them can ever overlap. Once this is granted, it is also clear that no individual can be said to exist in more than a world – that is, every individual will have to be ‘worldbound’. This, in turn, raises a problem. For if that is the case, how are we to make sense of the truth of (2)? The problem is that (2) can no longer be verified by a world where Obama lost the elections, since no world has Obama as a part of itself except for ours. Lewis’ famous solution to this problem is to appeal to the notion of counterpart. Since, strictly speaking, this is the only world where Obama exists, the truth-conditions for de re modal statements such as (2)25 must involve someone other than him; someone, for Lewis, who is one of Obama’s counterparts. Other worlds are thus viewed as populated by other-worldly people, none of whom is identical with anything existing outside of their world, but some of whom is a counterpart of other-worldly inhabitants (perhaps this-worldly ones). The counterpart relation can be defined in terms of resemblance in the following way: (C) an individual y in a world w’ is a counterpart of an individual x in a world w iff (i) y resembles x, and (ii) nothing else in w’ resembles x more than y. In the semantics that results from including the notion of counterpart and from denying transworld identity, de re modal statements receive translations that seek to capture what we find intuitively correct in them, without falling back into the simple analysis we started with. Thus, for instance, (2) would be analyzed as (3) ∃w (Ww Λ ∃x (Ixw Λ Cxo Λ Lx)), 25 The distinction between de re and de dicto modality is a distinction between the sorts of things that can be subjected to modal qualification. De re modal statements involve the ascription of a modal profile to an individual, whereas in de dicto modality modal properties are attributed to propositions. While both types of statements imply, within standard semantical treatments, a commitment to possible worlds, only de re modality implies a commitment to transworld identity. The latter is the claim that the same individual can exist in more than one possible world. 24 where ‘W’, ‘I’, ‘C’ and ‘L’ are predicates letters meaning ‘is a world’, ‘exists in’, ‘is a counterpart of’, and ‘lost the elections’ respectively, ‘x’ and ‘w’ are individual variables, ‘o’ is a singular term denoting Obama, ‘Λ’ the connective for conjunction, and ‘∃’ the existential quantifier. In words: there is a world in which there exists an individual x such that x is a counterpart of Obama and x lost the elections. Similarly, a sentence such as (4) Obama is necessarily human, would be rendered as (5) ∀w (Ww ⊃ ∀x((Ixw Λ Cxo) ⊃ Hx)), where, beyond the symbols used in (3), ‘H’ is a predicate letter that means ‘is human’, ‘⊃’ is the connective for the material conditional, and ‘∀’ the universal quantifier. In words: every counterpart of Obama in any world (where such there be) is human. This paraphrase strategy enables Lewis to stick to the intuition that nothing can exist in more than one world, while systematically verifying a sentence lying in the “neighborhood” of every de re modal statement that we intuitively regard as true. Now that we have seen some of the main features of concrete modal realism, we turn to one of its main competitors. 3.3.1.2 Abstract modal realism The driving impulse that guides abstract modal realism lies in the attempt to take hold of the benefits that come from quantifying over possible worlds without incurring a commitment to the generous ontology that characterizes concrete modal realism. As Lewis would put it, it is an attempt to get its advantages without paying the ontological prize that is due for them. On this view, possible worlds are ways – states, situations – things might be or might have been. The notion of a way things might be is unfolded by reference to abstract objects of some sort. Different views in this family can be distinguished depending on the kind of abstract object that is 25 appealed to, but for limitations of space we will focus only on the version that has been elaborated by Alvin Plantinga (1976). For Plantinga, states of affairs are among the ontological categories that should be taken as primitive. These are abstract entities, such as a’s being F or a’s standing in relation R with b, that consist of one or more particulars instantiating a property or standing in a relation, and that may or may not obtain. If a state of affairs obtains, the corresponding situation that things could be in is said to be actual – it is said to be part of the world we live in. A possible world is then characterized as a maximal state of affairs, and the notion of maximality (roughly, completeness) in turn is defined in terms of the logical notions of inclusion and preclusion. A state of affairs s is said to be maximal if, for every state of affairs a, either s precludes a or it includes a. And while a state of affairs s includes a iff s obtains only if a does, s precludes a iff s obtains only if a doesn’t. Then, possible worlds are accordingly viewed as maximal and possible states of affairs. As this definition makes clear, abstract modal realism does not seek or provide any reductive analysis of modality, since the notion of possibility is still explicitly appealed to in characterizing the states of affairs in terms of which possible worlds are explicated. Specifically, insofar as a possible world is regarded as a maximal state of affairs which is such that it possibly obtains, possible worlds are still explicated in thoroughly modal terms. Whether this feature of abstract modal realism is a cost or a benefit is a deep methodological issue, which depends on the value of reduction when it comes to specific modal matters. In this respect, the stances taken by the concretist and by the abstractionist are staunchly opposed: whereas the former takes concrete worlds as primitives and looks for ways to reduce modal notions to them, the latter has modal primitives and seeks to reduce possible worlds to those. We have seen that for the abstractionist the actual world is nothing but the total state of affairs that obtains, and that there is a possible world for any consistent state that the actual world might be in. Now, the abstractionist claims, holding these views comes with a crisp theoretical benefit. Because for him, unlike for Lewis, it is possible to take de re modal claims at face value 26 without having to produce any paraphrase thereof. Indeed, he need not countenance either the thesis that individuals are worldbound or the claim that no two worlds may overlap. This is so because when merely possible, non-actual, situations are thought of as abstracta, it is not problematic to think of them as containing the same objects – including actual ones. And indeed, given that transworld identity is consistent with this conceptual framework, many authors working within it have accepted it, viewing this as a major benefit of their theory. 4. Existence and validity So far we have seen that the kinematics of law-creation involves the performance of the specific type of illocutionary act that we called ‘law-making’, and sought to characterize the latter in dynamic terms as the act whose essential effect consists in updating the legal system. Furthermore we saw that, necessarily, every lawmaking act has content, and that this is the content which is added to the legal system when a lawmaking act is successfully performed. From here, we were naturally led to consider the main candidate conceptions of content, so as to review some of the hypotheses about what its nature could be. Now we turn to the last set of issues that will be examined in this paper. Recall that in our framework legal systems are nothing but sets of laws, and laws are contents that possess the property of being valid. Hence, the legal system of a jurisdiction j at t is the simply the set of the contents {p1, …, pn} that are valid in j at t. Given this setting, three central and closely related questions emerge, which force us to look into the nature of existence, and the relation between existence and validity: Q1 what is it for a content to exist? Q2 what is it for a law to exist? Q3 what is it for a content to be valid? 27 We start with Q1. Hans Kelsen (1945) famously held that validity is the mode of existence specific of norms. Let us assume, for ease of exposition, that norms (rules) are contents; 26 then, Kelsen’s contention turns into the claim that validity is the mode of existence specific to those contents that are norms. Arguably, this claim may be interpreted in at least two different ways: on one of them, he would be talking nonliterally, and simply saying that norms acquire something very special when they become valid; on another, he would be speaking literally, and thereby ascribing a distinctive way of existing to them. We shall cope with the second reading only. Thus interpreted, Kelsen’s claim implies that there are at least two different ways in which something might be said to exist; this in fact amounts to the view that ‘exist’ is ambiguous, as it would express different meanings when applied to different objects. The thesis that existence is equivocal has been advocated also by Gilbert Ryle (1949: 23), who writes: It is perfectly proper to say, in one logical tone of voice, that there exist minds and to say, in another logical tone of voice that there exist bodies. But these expressions do not indicate two different species of existence . . . . They indicate two different senses of ‘exist’, somewhat as ‘rising’ has different senses in ‘the tide is rising’, ‘hopes are rising’, and ‘the average age of death is rising’. As we can see, the thesis in question is not without intuitive byte, and in fact still has some advocates in contemporary metaphysics. Nevertheless, the view is also inconsistent with a central (nowadays dominant) meta-ontological tenet, ultimately traceable to the work of W.V.O. Quine (1948). Two fundamental theses on which the Quinean metaontology rests are (i) that being is the same as existence, and (ii) that existence (and, hence, being) is univocal. According to the former claim, to say (for instance) that cats exist is equivalent to saying that there are cats – or, better, that there is at least one of them, or that something is a cat. In a slogan, existence is being, and being is quantification: so there are no things that do not exist. Further, many theorists working in this tradition also think that a fruitful and faithful way of expressing ontological claims is 26 In case you worried about this stipulation, it is worth noticing that it is immaterial with respect to the argument we now present, and merely serves to connect it to the answer we shall later give to Q1. 28 provided by the language of first order logic: fruitful, since it enables metaphysicians to achieve clarity and rigor in conducting their debates; faithful, since the meaning of ‘exist’ is accurately captured by the existential quantifier ‘∃’. Accordingly, such theorists would regard as ideal to regiment the language of ontology through the means offered by formal logic; a characteristically ontological contention like ‘propositions exist’, would be taken to mean the same as ‘something is a proposition’, which in turn could be expressed by (6) ∃x (Px) where ‘P’ is a predicate letter meaning ‘is a proposition’. The second thesis – that existence is univocal – has been forcefully advocated by Van Inwagen (2002). Whereas philosophers like Ryle and Kelsen were drawn to the thought that different kinds of things could enjoy different ways of being, Van Inwagen and Quine stick to the simple idea that what may appear to be differences in being are, in fact, no more than differences in kind. So, for instance, beyond the uncontroversial fact that chairs and numbers, if there are any, are entities of quite different sorts, this is not because they exist differently, but simply because they are different. That this is so, Van Inwagen thinks, can be appreciated by noticing the similarity between numerals and expressions such as ‘there is’ and ‘exist’. First, it seems clear that number words like ‘one’ or ‘five’ are unambiguous, since it is in the ‘very essence of the applicability of arithmetic’ that they can be used to count just about anything. So if José Juan has three pencils and Sam has three dogs, the number of José Juan’s pencils is the number of Sam’s dogs. And intuitively, being and numbers are also closely related: for to say that there are no dogs is equivalent to saying that their number is zero, while to say that there is a dog is equivalent to saying that the number of dogs is one or more. So if numerals are not ambiguous, ‘exist’ is not either.27 Eventually we come to be in a position to answer Q1: for a content to exist just is for there to be (at least) one such thing. 27 The argument is readapted from Van Inwagen (1998). 29 Let us move on to Q2. Assuming the meta-ontological framework we just sketched, coupled with the analysis of law that we gave, we might say that for a law to exist is for there to be a valid content. This, in fact, is equivalent to saying that laws exist iff there are any contents that instantiate the property of legal validity. Thus, on the view we advocate, existence and validity should be kept distinct, since a content may exist without being valid – as indeed many do. Think, for instance, of the content that one must assert only what one knows; this may well be the norm that governs assertion and, whether it governs it or not, it exists as an abstract entity irrespectively of whether you or us ever happen to think of it. And yet, since no legal system (at least none we are aware of) has it as one of its members, the content is not valid. The same could probably be said of the content that one ought to kill one’s neighbor, and of many others. Validity is possessed only by a small subset of all the available contents – by those, in particular, that are so endowed in virtue of the obtaining of certain facts, which is the aim of jurisprudence to uncover. This reflection brings us to the last of our questions. Q3. What is it for a content to be valid? A possible answer to this question might be that a content is valid in a jurisdiction j just in case it is a member of j’s system. Although saying this can be helpful, since it clarifies our subject matter by making clear what sort of validity it is that we are singling out and talking about, since defining ‘validity’ in terms of membership to legal systems would be circular, something else must be said. The real question then becomes: what does it take for a content to become part of a legal system? For a content to be law is for it to have the property of legal validity; as we saw, drawing a distinction between a content’s existence and its instantiation of validity allows one to hold that a large number of contents exist while regarding only a subset of them as being part of some jurisdiction’s system. One benefit of this conceptual framework is that it enables us to separate the aspect of law that makes it an abstract and mind independent object, from that which makes its existence rest on the actions and intentional states of humans, while recognizing both of them. One can explain the sense in which laws are abstracta (they cannot be perceived, touched, heard), by appealing to the fact that they 30 are essentially contents, and account for their social dimension by appealing to the fact that whether a given content is valid depends on a number of other things, including facts about human activities such as drafting and voting, the existence of legal institutions like chambers and governments, the performance of certain speech acts, social conventions, or conformity to moral principles. The accurate identification of the variety of facts that make it the case that the law is what it is no doubt constitutes a daunting task, and one that goes widely beyond the scope and purpose of this paper. So we will not provide an answer to Q3. However, Q3, as embedded in the theoretical framework that we have sought to delineate, offers a useful and precise way of framing the central issue for the philosophy of law. Thus interpreted, the question is: what are those facts in virtue of which contents have the property of being valid? This question is one for legal theorists to tackle, and the positivism-antipositivism debate will thus continue to unfold, through the defense of their rival views on this matter.28 Still, we hope to have made a good case for thinking that the setting in which this inquiry should be pursued, along with its ontological and metaphysical features, is one that carries independent theoretical interest, and one that has a crucial role to play vis-à-vis the progressive development of the inquiry itself. 5. References Alchourron, C.E., & Bulygin, E. (1981) ‘The expressive conception of norms’, in Hilpinen, R. ed. (1981). Anscombe, G.E.M. (1957) Intention, Basil Blackwell, Oxford. Austin, J.L. (1962) How to do things with words, OUP, London. Bach, K., & Harnish, R.M. (1979) Linguistic communication and speech acts, MIT University Press, Cambridge Massachussets. 28 Of the extended literature on the subject, we indicate Hart (1994), Dworkin (1986), Raz (1979), and Greenberg (2004). 31 Bianchi, C. ed. (2004) The semantics/pragmatics distinction, CSLI Lecture Notes, University of Chicago Press, Stanford. Brown, J., & Cappelen, H. eds. (2011) Assertion, OUP, Oxford. Carnap, R. (1947) Meaning and necessity, University of Chicago Press, Chicago. Davidson, D., & Harman, G. eds. (1970) Semantics of natural language, Synthese library. Dworkin, R. (1986) Law’s empire, Harvard University Press. Frege, G. (1891) ‘Funktion und begriff’, in Hermann Pohle. (1892/1960) ‘Über Sinn und Bedeutung’, in 100 (1) Zeitschrift für Philosophie und Philosophische Kritik 25. (1980) Philosophical and mathematical correspondence, University of Chicago Press, Chicago. Garcia-Carpintero, M. (2004) ‘Assertion and the semantics of force-markers’, in Bianchi (ed.), 133-166. Green, M. (2015) ‘Speech acts’, The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/sum2015/entries/speech-acts/>. Greenberg, M. (2004) ‘How facts make law’, in 10 Legal Theory 157. Grice, H.P. (1989) Studies in the way of words, Harvard University Press, London. Hanks, P. (2007) ‘The content-force distinction’, in 134 (2) Philosophical Studies 141. Harris, D. (2014) Speech act theoretic semantics, PhD dissertation, City University of New York. Hart, H.L.A. (1994) The concept of law, Oxford University Press, Oxford. Heim, I., & Kratzer, A (1998) Semantics in generative grammar, Blackwell, Oxford. Hilpinen, R. ed. (1981) New studies in deontic logic, Dordrecht, Reidel. Hintikka, J. (1957) ‘Quantifiers in deontic logic’, in 23 (4) Societas Scientiarum Fennica, Commentationes humanarum litterarum, Helsinki. Kelsen, H. (1945) General theory of law and state, trans. A. Wedberg, Russell and Russell, New York. King, J.C., Soames, S., & Speaks, J. (2014) New thinking about propositions, OUP, Oxford. 32 Kratzer, A. (1977) ‘What ‘must’ and ‘can’ must and can mean’, Linguistics and Philosophy 1: 337–55. Kripke, S.A. (1959) ‘A completeness theorem in modal logic’, in 24 (1) Journal of Symbolic Logic 1. Lewis, D.K. (1970) ‘General semantics’, in 22 (1-2) Synthese 18. (1979a) ‘A problem about permission’, in Saarinen, E., Hilpinen, R., Niiniluoto, I., & Provence, M. eds. (1979), 163-175. (1979b) ‘Scorekeeping in a language game’, in 8 (1) Journal of Philosophical Logic 339. (1986) On the plurality of worlds, Blackwell Publishers. MacFarlane, J. (2011) ‘What is assertion?’, in Brown, J., & Cappelen, H. eds. (2011). Marcus, R.B. (1946) ‘A functional calculus of first order based on strict implication’, 11 Journal of Symbolic Logic 1. Menzel, C. (2015) ‘Possible worlds’, in The Stanford Encyclopedia of Philosophy, Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/sum2015/entries/possible-worlds/>. Midgley, G.C.J. (1959) ‘Linguistic rules’, in 59 Proceedings of the Aristotelian Society, New Series 271. Montague, R. (1970) ‘Pragmatics and intensional logic’, in 22 (1-2) Synthese 68. Plantinga, A. (1976) ‘Actualism and possible worlds’, in 42 (1-3) Theoria 139. Portner, P. (2007) ‘Imperatives and modals’, in 15 Natural Language Semantics 351. Quine, W.V.O. (1939) ‘Designation and Existence’, in 36 Journal of Philosophy 701. (1948) ‘On what there is’, in 2 (5) Review of Metaphysics 21. Rawls, J. (1955) ‘Two concepts of rules’, in 64 (1) Philosophical Review 3. Raz, J. (1979) The authority of law: essays on law and morality, OUP, Oxford. Russell, B. (1903) Principles of mathematics, Routledge, London and New York. Ryle, G. (1949) The concept of mind, Hutchinson, London. 33 Saarinen, E., Hilpinen, R., Niiniluoto, I., & Provence, M. eds. (1979) Essays in honour of Jaakko Hintikka on the occasion of his fiftieth birthday on January 12, 1979, Reidel. Searle, J.R. (1969) Speech acts: an essay in the philosophy of language, Cambridge University Press. (1985) Expression and meaning, Cambridge University Press. Stalnaker, R. (1970) ‘Pragmatics’, in Davidson, D., & Harman, G. eds. (1970). (1978) ‘Assertion’, in 9 Syntax and Semantics 315. (1999) Context and content: essays on intentionality in speech and thought, Oxford University Press, Oxford. Strawson, P. (1964) ‘Intention and convention in speech acts’, in 73 (4) Philosophical Review 439. Van Inwagen, P. (1998) ‘Meta-ontology’, in 48 Erkenntnis 233. (2002) ‘The number of things’, in 12 (1) Philosophical Issues 176. Weinberger, O. (1985) ‘The expressive conception of norms – an impasse for the logic of norms’, in 4 (2) Law and Philosophy 165. Williamson, T. (1996) ‘Knowing and asserting’,105 (4) Philosophical Review 489. 34